14th Congress of the International Society of Photogrammetry,
Hamburg 1980
Commission VII
Working Group VII, 4:
Oceanography and Sea Ice
Invited Paper
R. C. Beal
The Johns Hopkins University
Applied Physics Laboratory
Johns Hopkins Road
Laurel, Maryland 20810
DETECTION AND TRACKING OF A LOW ENERGY SWELL SYSTEM
OFF THE U.S. EAST COAST WITH THE SEASAT SAR
ABSTRACT
On the morning of 28 September 1978, at 1520 GMT,
SEASAT approached the east coast of the United States, with
the 100 km swath of its synthetic aperture radar (SAR) running
approximately parallel to the coast, but displaced eastward by
about 20 km.
On the basis of the present analysis of that
pass, the following major conclusions may be stated:
(a) the
SAR can successfully detect low-energy swell systems with significant wave heights, H , of well under 1m (actually 0.65 ±
s
0.25 m); (b) the refraction of low-energy but well-organized
swell resulting from changes in the local depth of the ocean
is clearly detectable in both wavelength and direction; and
(c) the complexity of the ocean spectrum (e.g., whether it is
composed of more than one system or is spread in direction and
wave number) seems to have little bearing on the threshold
detection limits.
BACKGROUND
Studies made with aircraft in the past few years have
indicated that ocean swell can be imaged with a SAR, at least
for some wind velocities when there is a substantial component
of the swell traveling along the radar line of sight.
However,
the bounds of wind, wave, and geometric conditions for which the
detection of ocean waves is reliable remain elusive for want of
an extensive experimental data base.
SEASAT provided a unique,
although limited, opportunity to re-examine the wave detection
problem with few of the artificial constraints of aircraft measurements.
Jordan (1978) gives a concise summary of the SEASAT
SAR design parameters; Beal (1978) gives a more general description of the system and of its fundamental limitations.
Pre-
liminary assessments of its ocean wave detection capabilities
have been compiled by Gonzalez (1979) and Beal (1979,1980).
During the 100 day lifetime of the SEASAT SAR, nearly
500 passes of l to 15 min duration were collected at three domestic and two foreign receiving stations.
Twenty-three passes
provided acceptable SAR imagery within 70 km of a well instrumented "sea truth'' pier operated by the U.S. Army Corps of
Engineers, Coastal Engineering Research Center (CERC) at Duck,
North Carolina.
APL and several government agencies collected
wind and wave measurements from 12 August to 9 October 1978.
DISCUSSION
Figure l shows the areas for which estimates or measurements were collected on 28 September.
For reference, the
boundaries of the 100 km SAR swath are shown by the solid lines
inclined approximately 25° with respect to north at this latitude.
The locations are keyed alphabetically from north to
south .
Position A corresponds to the Navy Fleet Numerical
Weather Center (FNWC) grid point 271. Positions B, C, E, and
F are 15 km 2 areas over which the SEASAT SAR imagery was optically Fourier transformed.
The appreciable changes in local
ocean depth (dotted contours in Fig. l) at B, C, and E would
cause spreading of a single frequency deep-water wave in both
wavelength and direction .
Deep water dispersion prevails only
at F for wavelengths greater than 70 m.
075.
Local depth changes
at B are especially severe, ranging from less than 10 to at
least 20m.
The laser profilometer spectrum in Fig. 2 was
collected at C by a National Oceanic and Atmospheric Administration/Sea Air Interaction Laboratory (NOAA/SAIL) aircraft.
In-situ one-dimensional spectral measurements were collected
at the CERC pier (D).
The FNWC grid point 260 (G) provides a
convenient reference spectrum for the SAR imagery data collected at F.
Fig. 1.
F and G are the only deep water locations of
Time histories of the long (greater than 1 s) waves
were recorded by five instruments in the vicinity of the research pier for a 20 min interval spanning the satellite overpass time.
The one-dimensional long-wave spectrum was measured
near the pier with two Baylor gauges, two wave-rider buoys, and
one capacitive wave staff.
Spectra from each instrument show
varying amounts of 11 and 7 s systems .
The average of the two
gauges and the two buoys (Fig . 2) clearly identifies each wave
system.
The significant wave height derived from the averaged
spectrum is 1.0 m, which of course includes the combined energy
from both the 11 and the 7 s systems .
The significant wave
height for each separate system (roughly equal in energy at
the pier) is closer to 0 . 7 m.
By comparison, the FNWC grid
point closest to the pier (G) yields estimates of 0 . 4 and 0 . 6
m for the long-wave and the short-wave systems, respectively.
The NOAA aircraft flying about 50 km to the east of
the pier was equipped with a laser profilometer to measure the
one-dimensional ocean height spectrum in the flight direction
of the aircraft.
For a typical "swell-run", the aircraft head -
ing is chosen to correspond to the vector direction of the swell
(within a few degrees) .
The resulting spectrum accurately mea-
sures the total energy on the surface if the spectrum contains
little energy orthogonal to the flight vector because that energy is effectively excluded .
The laser spectrum plotted a -
gainst the averaged pier spectrum in Fig. 2 verifies a double
wave system with predominant periods at 11 and 7 s, and a total
significant wave height of 1.26 m, or approximately 0 . 9 m for
each system.
The energy of the 7 s system is probably under-
076.
estimated because of the directional sensitivity of the laser
profilometer .
However , the energy of the 11 s system should
be accurate and probably represents an effective upper bound
to the significant wave height in the transformed areas.
The SAR was activated for approximately four minutes
on 28 September as it approached the U.S. East Coast .
The
radar interaction wavelength is about 30 em , and the image in tensity (or reflected power) is generally proportional to the
amplitude of Bragg scatterers of the 30 em wavelength on the
ocean surface (Wright , 1968) .
The amplitude of the scat -
terers may, in turn , be strongly (but not solely) correlated
with surface wind at the boundary layer.
In general , brighter
regions in an image correspond to higher winds and darker regions to lower winds .
The imagery from the SAR pass on 28 September was op tically Fourier transformed , d i gitally scanned at the equivalent
of 6 m ground resolution , and spatially averaged using a 7 x 7
element sliding window .
The contrast was enhanced by means of a
three - segment piecewise linear leve l transformation with a very
high gain in the center segment .
Each set of break po i nts was
individually optimized to compensate for the variations in aver age intensity .
Figure 3 shows a progression of enhanced optical Four ier transforms representing the image spectra at F, E, C , and
B as the wave trains approach shore .
The sequence shows quite
clearly the refraction effects of the variations in ocean depth
on the long-wave component .
A 210 m wave in deep water at F
shortens to 170 mat E , to 160 mat C , and finally to 120 m at
B, in shallow water .
Furthermore , the deep water spectrum of
F correlates well with the FNWC estimate made at G (grid point
260) of spectral peaks in both wave number and direction .
The
presence of a short wave sys t em is also evident on the trans forms .
This short wave energy correlat i on may be , at least in
part , an artifact of the ground processor.
Recent digitally
processed and transformed imagery of the same area shows much
weaker correlations in the short wave (7 s) portions of the
spectrum .
077.
It should be emphasized that no correlation between
spectral energy dens i ty and image transform density has been
attempted.
A proper treatment of that question would require
careful accounting of the many system nonlinearities , some
deliberate l y introduced for en hancement and some unknown .
A
better understanding of ocean backscatter models is also a
prerequisite for further progress here .
The quantitative shallow- water dispersion re l ation ship (Kinsman , 1965) is shown by the curves in Fig . 4 , on which
are also plotted the results of Fig . 3 .
Each of the four cen -
ter wavelengths from the optical transforms is entered on the
horizontal ax i s of the dispers i on relationship and is trans ferred according to the variation i n depth for its particular
location .
The location of peaks from the optical transforms is
judged to be accurate to +5% .
These two uncertainties combine
to produce the areas of uncertainty shown in Fig . 4 .
The data
set is seen to be most consistent with the assumption of an
11 . 7 s swell system .
In summary, a well - organized, very - low - energy swell
system off the east coast has been tracked with the SEASAT SAR
from deep water, across the continental shelf, and into shallow
water .
The results indicate that a spaceborne imaging radar
can measure ocean wavelength and direction accurately, even in
coastal areas and in the presence of a mixed ocean .
For sepa -
rating swell systems in a mixed ocean , its accuracy may exceed
that of any other known technique .
ACKNOWLEDGMENTS
This work was jointly supported by the National
Oceanic and Atmospheric Administration and the National Aero nautics and Space Administration as a portion of the SEASAT
Announcement of Opportunity Program under Contract MO-AOl - 78 00 - 4330 .
The laser spectrum of Fig . 2 was provided by D . Ross
of NOAA .
The gauge spectrum of Fig . 2 was provided by D . Lichy
of CERC .
J. Jenkins of APL enhanced the images shown in Fig. 3 .
078.
REFERENCES
1.
R . Jordan , " The SEASAT-A Synthetic Aperture Radar Design
and Implementation", Proc . Synthetic Aperture Radar
Technology Conference, Las Cruces, N.M. , 1978.
2.
R . C . Beal , "Useful Spaceborne Synthetic Aperture Radars ",
Proc . XXIXth International Astronautical Federation Con gress, Dubrovnik , Yugoslavia , 1978.
3.
F . I. Gonzalez et a1 ., "SEASAT Synthetic Aperture Radar:
Ocean Wave Detection Capabilities", Science 204, 1979,
pp. 1418-1421.
4.
R . C. Beal, The SEASAT SAR Wind and Ocean Wave Monitoring
Capabilities , APL/JHU SlR-79-U-019 , August 1979.
5.
R . C. Beal, "Spaceborne Imaging R<1dar Ocean Wave Monitoring", Science, Spring 1980.
6.
J. W. Wright, "A New
~1odel
for Sea Clutter", IEEE Trans.
Antennas and Propagation AP-16, No. 2, 1968, pp . 217-223 .
7.
B. Kinsman, Wind Waves, Prentice Hall, 1965, pp. 126-133.
079.
'
''
\. ..
,, ... "
I
'
Fig . 1 Locations of surface, aircraft, and spacecraft
measurements of the low- energy swell systems present on
28 September 1978 .
080.
- - - Ave rage of fou r spectral measurements in
vic inity of D (Hs = 1.0 m)
- - - NOAA ai rcraft laser prof il ometer spectrum in vicinity of C ( Hs = 1.26 m)
1.5
11
<J)
N
E
.£ 1.0
<J)
c
a.>
"0
a.>
(.)
c
.~
.... 0 .5
<U
>
0~------~-------L------~--------~--~
0 .20
0.15
0.10
0
0.05
Frequency (Hz)
Fig . 2 A comparison of surface and air measurements of the
wave height spectra near Duck , North Carolina .
DB::t.
FNWC OVERLAY
GRID POINT 260
ILOCATION G)
N
LOCATION@]
INTERMEDIATE DEPTH
~ 20m
N
w
w
Fig. 3 Optically processed , optically transformed , and
digitally enhanced SAR wave spectra from each of the four
locations F, E, C, and B. The sequence moves from deep to
shallow water and illustrates both wavelength and direction
change as the 11 s swell system approaches shore . The overlap
of the FNWC spectra for grid point 260 (G) on the SAR image
spectrum at F shows excellent correlation of the 11 s swell
system present on 28 September 1978 . Overlay units are
expressed in relative energy units per cell.
082.
Note : Uncerta inty areas are due to changes in bathymetry over transform area of 15 km2 and to peak
wavelength error of ±5% of nominal.
Loca l depth (m)
2
5
10
20
40
(J)
15
"'tJ
0
1....
Q)
Q.
Cll
>
C'O
10
~
c
C'O
Cll
(.)
0
5
,,I I
I Most
II
I
El IE I E
§ II~
0
probable peri od:
j11.7 s or 0.085 Hz
IS
N
L-------~~--~--~------~------~
0
100
200
300
400
Ocean wavelength (m)
Fig . 4 The results of Fig . 3 plotted on the shallow-wa t er
dispersion relationship.
083.